Neutral wind influence on the electrodynamic coupling between the ionosphere and the magnetosphere

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 7, NO. A,,.9/JA9, Neutral wind influence on the electrodynamic coupling between the ionosphere and the magnetosphere C. Peymirat Centre d Etude Spatiale des Rayonnements, Toulouse, France A. D. Richmond and R. G. Roble High Altitude Observatory, National Center for Atmospheric Research, Boulder, Colorado, USA Received March ; revised May ; accepted June ; published January. [] The influence of the ionospheric wind dynamo on the steady state electrodynamic interaction between the ionosphere and the inner magnetosphere is evaluated with the Magnetosphere-Thermosphere-Ionosphere Electrodynamics General Circulation Model (MTIE- GCM) of Peymirat et al. [99]. Two types of interaction between the solar wind magnetosphere (SWM) dynamo and the ionospheric dynamo are considered: either the SWM dynamo is a voltage generator imposing a fixed electric potential in the polar cap or it is a hybrid current/voltage generator such that the electric potential in the polar cap results from the combined action of the SWM and ionospheric dynamos. The following results are found. () The dynamo effect of winds that are accelerated by the high-latitude ion convection increases the meridional electric field in the auroral zone but reduces the potential variation around the Auroral Zone Equatorial boundary (AZEQ) and the plasma pressure in the nightside magnetosphere close to the Earth, corresponding to an enhancement of the shielding effect produced by Region field-aligned currents. The magnitudes of the wind-induced effects are on the order of % for the shielding potential and % for the plasma pressure reduction. () The wind-induced reduction of magnetospheric plasma pressure near the Earth is times larger for a doubling of the plasma source density in the tail. () The relative influence of winds on shielding is similar for polar cap potential drops of kv and 7 kv. () When the wind is allowed to influence the polar cap potential, it increases the potential drop along the polar cap boundary by % but does not modify the net influence on the shielding potential along AZEQ. (5) The influence of the neutral winds set up by the high-latitude convection is rest ricted to the aurora l zone. I NDEX TERMS: 7 Magnetospheric Physics: Magnetosphere ionosphere interactions, 7 Magnetospheric Physics: Magnetosphere inner, 7 Magnetospheric Physics: Electric fields; KEYWORDS: Magnetosphere, ionosphere, thermosphere, neutral winds, coupling. Introduction Copyright by the American Geophysical Union. -7//JA9 [] The magnetosphere-ionosphere system has multiple sources of electric fields and currents. These fields and currents tend to be readily transmitted throughout the system, producing mutual electrodynamic interactions among the sources. For the purposes of the present paper, it will be convenient to consider the three dominant electrodynamic sources as the solar wind magnetosphere (SWM) dynamo, the ionospheric dynamo, and the magnetospheric thermoelectric (MT) dynamo. The SWM dynamo is induced by the interaction of the solar wind and the geomagnetic field, generating an electric field which is transmitted to the polar ionosphere via Region field-aligned currents that connect the solar wind currents to the polar cap ionospheric currents. The ionospheric dynamo is induced by neutral winds that move the conducting ionosphere through the geomagnetic field. Collisional drag by the convecting ions tends to give the wind velocity a component in the direction of the ion velocity. The MT dynamo arises from gradient-curvature drifts of energetic particles on closed magnetic field lines. These particles carry current across the field lines both through the gradient-curvature drift and through their gyromotion, and this cross-field current diverges and feeds into the Region fieldaligned currents. Viewed differently, the cross-field currents are those necessary to produce a force balance with pressure gradients of the magnetospheric plasma. The object of this paper is to explore the manner in which interactions among these three electrodynamic sources can influence the resultant distributions of electric fields and currents in the ionosphere and the distribution of plasma pressure in the magnetosphere. [] The influence of the ionospheric dynamo on magnetosphere/ionosphere electrodynamics has often been studied from two simplified points of view: treating the solar wind magnetosphere sources either as a voltage generator or as a current generator. Specifying the electric potential while allowing the field-aligned currents to vary in response to the ionospheric dynamo corresponds to the voltage generator, while the reverse corresponds to the current generator. Using the voltage generator concept, Axford and Hines [9], Lyons et al. [95], Deng et al. [99, 99], Lu et al. [995], and Peymirat et al. [99] illustrated how the neutral winds tend to generate currents which oppose the currents driven by the external sources. Using the current-generator concept, Banks [97], Richmond and Matsushita [975], Deng et al. [99], and Richmond [995b] showed how the neutral winds tend to create a potential pattern similar to the potential pattern due to the magnetospheric sources. Effectively, the winds tend to entrain the ions by creating an electric field which produces an electrodynamic drift in the wind direction. This can increase the overall strength of magnetospheric convection and SMP -

2 SMP - PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING may allow the magnetospheric plasma to penetrate closer to the Earth, as suggested by Peymirat et al. [99]. [] The MT dynamo is associated with the Region fieldaligned currents created during the convection of the magnetospheric plasma. In a steady state the effects of these currents tend to reinforce the north-south or meridional component of the electric field in the auroral region while decreasing both components of the electric field equatorward of the auroral region [Fejer, 9; Block, 9; Vasyliunas, 97, 97], creating the so-called shielding phenomenon. The reduction in the westward field on the nightside prevents the magnetospheric plasma from penetrating more closely to the Earth. Wolf et al. [9], Spiro et al. [9], Fejer et al. [99], and Forbes and Harel [99] used the Rice Convection Model (RCM) [Harel et al., 9] to study the influence of the neutral winds on the magnetospheric convection. Wolf et al. [9] showed that the effect of neutral winds driven only by solar thermal and tidal forcing is negligible for the distribution of magnetospheric plasma. During the recovery of a storm, Spiro et al. [9] and Fejer et al. [99] illustrated how the neutral winds tend to maintain the convection pattern previously set up during the main phase of the storm. For steady state active magnetic conditions, Forbes and Harel [99] presented results illustrating how neutral winds can increase the effectiveness of shielding. Similar conclusions were obtained by Peymirat et al. [99] with the Magnetosphere-Thermosphere-Ionosphere Electrodynamics General Circulation Model (MTIE-GCM), which calculates in a more consistent way the ionospheric dynamo influence. They assumed an SWM voltage generator over the polar cap but allowed for a realistic interaction between the MT dynamo and the ionospheric dynamo equatorward of the polar cap boundary (PCB). They pointed out that neutral winds do increase the shielding but have only a minor influence on the Region field-aligned currents associated with the MT dynamo, which behaves largely like a current generator. They hypothesized that by allowing the winds to influence the potential over the polar cap, which was not allowed in their simulations, larger east-west electric fields could be created around the PCB and in the auroral zone, fields that might produce a net wind-induced increase rather than decrease of the east-west electric field equatorward of the auroral region. If so, the net effect of winds on shielding might be to decrease rather than increase it. [5] In this study we use the MTIE-GCM to extend the SWM voltage generator study of Peymirat et al. [99], and we begin to model the influence of neutral winds on magnetosphere-ionosphere coupling for a case where the SWM dynamo interacts with the polar cap in a manner intermediate between a voltage generator and a current generator. Both studies are performed for steady states of the magnetosphere, ionosphere, and thermosphere system. Section describes the model and the input parameters of the model. Section analyzes the SWM voltage generator case to evaluate how neutral winds modify the ionospheric electric field equatorward of the PCB and how the interaction between the MT dynamo and the ionospheric dynamo varies as a function of the magnetospheric plasma distribution and the polar cap potential drop. Section introduces a hybrid SWM current/voltage generator, in which the potential at and inside the PCB can be altered by the wind. We test the hypothesis of Peymirat et al. [99] that allowing winds to change the potential at the PCB might cause winds to decrease rather than increase the effective shielding of high-latitude electric fields from lower latitudes.. Model and Simulation Conditions.. Model [] The MTIE-GCM corresponds to the Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM) of Richmond et al. [99] coupled with the Ionosphere-Magnetosphere Model (IMM) of Peymirat and Fontaine [99]. It has the capability to simulate self-consistently the dynamics of the magnetosphere, the ionosphere, and the thermosphere in regions where the magnetic field lines are closed. It computes the two-dimensional convection of the magnetospheric plasma in the equatorial plane of the magnetosphere and the three-dimensional structure of the density and the temperature of the ionosphere and the thermosphere above 97 km. The couplings among these three regions due to the auroral precipitation, the Region field-aligned currents, and the mapping of the electric field between the ionosphere and the magnetosphere are selfconsistently computed. Special emphasis is placed on the electric dynamo which takes into account auroral precipitation, the MT dynamo associated with the Region field-aligned currents, and the ionospheric dynamo induced by the neutral winds. As the model is based on a dipole magnetic field in the magnetosphere, it cannot exactly reproduce reality, but it can aid in the understanding of the complex couplings among the thermosphere, the ionosphere, and the magnetosphere. Another limitation of the current version of the model is its neglect of plasma heat flux across magnetic field lines, which Heinemann [999] showed can modify the velocity of information transfer in the magnetospheric plasma by %. The required inputs to the MTIE-GCM are the solar flux, upward propagating tides at 97 km, the polar cap precipitation, the polar cap convection pattern, and the density and temperature of the magnetospheric plasma in the tail source... Polar Cap Potential Pattern [7] The PCB is modeled as a circle centered on the magnetic pole. We set its radius to., which corresponds to a latitude of the MTIE-GCM grid, as in the work by Peymirat et al. [99]. The PCB radius is held fixed during the simulations. For our SWM voltage-generator simulations the electric potential inside the polar cap is specified from one of three different models, while for our SWM hybrid current/voltage generator simulation the potential inside the polar cap is calculated in a manner described in section, allowing for feedback from the ionospheric dynamo. The SWM voltage-generator cases in section use the potential model of Heelis et al. [9], with a sinusoidal shape of the potential along the PCB that Boyle et al. [997] showed to be adequate for southward steady states of the interplanetary magnetic field (IMF) and with a polar cap potential drop of either kv or 7 kv. Although the polar cap radius of. is consistent with a polar cap potential drop of 7 kv, it overestimates the size of the polar cap for a potential drop of kv, where a radius of 5 is expected [Peymirat and Fontaine, 997]. However, keeping the PCB fixed allows us to distinguish the effects associated with variations of the polar potential drop from those related to a displaced PCB. The SWM voltage-generator case in section has a different distribution of potential over the polar cap, which is matched to the distribution that would be produced by the hybrid current/voltage generator described in that section if there were no thermospheric winds. This allows a direct comparison between the voltage-generator and the hybrid current/voltage-generator models. The total potential drop is kv, and the potential maximum and minimum lie poleward of the PCB... Magnetospheric Source [] The magnetospheric source of the MTIE-GCM is located in the plasma sheet, beyond the external boundary of the domain where the model calculates the magnetospheric convection. This boundary is set to. Earth radii (R E ), i.e., the equatorial conjugate of the PCB at 7. magnetic latitude. [9] Statistical observations of the ion and electron populations show that the average values of the density range from.5 to cm, and the ion temperature ranges from kev to.7 kev [Huang and Frank, 9; Huang et al., 99; Baumjohann and Paschmann, 99; Baumjohann et al., 99; Goertz and Baumjohann, 99; Huang and Frank, 99; Escoubet et al., 997]. We set the ion temperature of the source to 5 kev. The electron temperature should have a value of /(7. 7.) the ion temperature

3 PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING SMP - [Baumjohann and Paschmann, 99], that is,. kev. However, as noted by Peymirat et al. [99], a larger value of kev is required for numerical stability of the ionospheric calculations in the version of the MTIE-GCM that we use for the present study. The density is set to. cm. [] Schumaker et al. [99] showed that the electrons precipitate with roughly 5% the rate computed under the strong pitch angle diffusion assumption. As the temperature of our magnetospheric electron source is rather large, we adjusted the pitch angle diffusion rate in the IMM to provide electron precipitation in agreement with the statistical observations of Hardy et al. [95]. We found that a rate of % the strong pitch angle diffusion is in rough agreement with the observations, i.e., the model reproduces the maximum observed energy fluxes. Similarly, to be consistent with the observed ion precipitation [Hardy et al., 99], the model ion precipitation is computed at % the strong pitch angle diffusion rate... Simulation Conditions [] We run the model for solar maximum equinox conditions corresponding to F.7 = Wm Hz, similar to Peymirat et al. [99]. Upward propagating atmospheric tides and particle precipitation in the polar cap are disregarded. The initial state of the MTIE-GCM results from the combination of two initial states independently calculated with the TIE-GCM and the IMM, as done by Peymirat et al. [99], using a polar cap potential drop of kv. The coupled models are run for one additional hour at kv to allow them to adjust to each other and to provide the initial state of the MTIE-GCM for the various simulations described below. [] Each simulation is run for hours under steady state forcing. For each fully coupled simulation, an additional simulation is performed in which the ionospheric dynamo due to the neutral winds is disregarded, similar to Peymirat et al. [99]. These simulation pairs are referred to hereafter as Run W, for Run With Winds, and Run NW, for Run With No Winds. Differences between the two runs help identify the role of the winds on the coupled electrodynamics.. Analysis of Magnetosphere-Ionosphere- Thermosphere Interactions for a SWM Voltage Generator [] For the simulations described in this section, we use as done by Peymirat et al. [99] the potential model of Heelis et al. [9] inside the polar cap. The results are displayed in Figure for a polar cap potential drop of kv and a source density of. cm. Shown are the total plasma pressure in the equatorial plane of the magnetosphere for Run W, the pressure difference between Run W and Run NW, the total potential in the ionosphere above magnetic latitude for Run W, the potential difference between Run W and Run NW, and the neutral winds at an altitude of 5 km in the Northern Hemisphere. This altitude is chosen because the corresponding neutral winds are representative of the ion-dragaccelerated winds which drive the ionospheric dynamo. [] Figure is similar to Figure of Peymirat et al. [99], which was obtained for half the source density but was run for hours instead of hours as done here. The close connection between the pressure and potential differences has already been discussed by Peymirat et al., [99] and will not be repeated here. The pressure distribution displays a peak of ppa, and the difference pressure distribution displays a negative peak of about 55 ppa, which yields a measure of the wind influence of about 55 ¼ : a little more than four times larger than found by Peymirat et al. [99]. This is associated with a relative decrease of the total potential variation around the Auroral Zone Equatorial boundary (AZEQ), which we take to be represented by. latitude, of about :þ: ¼ : which is similar to the value of. that can be deduced from the results of Peymirat et al. [99]. Obviously, the response is strongly nonlinear. [5] In order to reveal the nature of the wind effects in more detail, we examine the components of the electric field. Figure shows the latitude variations of the eastward and southward components of the electric field for different runs at five different magnetic local times (MLT) on the nightside. Strictly speaking, what is shown are the scaled eastward and negative northward electric field components, E mf and E ml, respectively, in the modified-apex coordinate system [Richmond, 995a], where E mf Rcosl m E ml m and where is the electric potential, l m and f m are the modified apex latitude and longitude, and R is a reference radius ( km, corresponding to a shell at km altitude). The solid lines correspond to Run W (including wind dynamo effects), and the dotted lines correspond to Run NW (without wind dynamo effects). The dashed lines are for an additional run that includes wind dynamo effects above AZEQ but not at lower latitudes. This helps isolate the influence of the high-latitude winds. Between and MLT a comparison of eastward electric fields when high-latitude wind dynamo effects are included (dashed and solid curves) with those when all wind dynamo effects are excluded (dotted curves) reveals that the neutral winds make the field around AZEQ less westward. Furthermore, a comparison of southward electric fields at dawn and dusk for the different runs reveals that the winds increase the strength of the north-south field in the auroral region. Both of these changes are characteristic of increased shielding. Comparing the dashed and dotted curves, one recognizes that the influence of the neutral winds is mostly limited to the high-latitude region, although some penetration of the eastward field up to in latitude is seen between and MLT. Comparing the solid and dashed curves, one recognizes that the influence of low-latitude and midlatitude winds is sometimes as important as that of the highlatitude winds around AZEQ. As we are interested in studying the effect of the neutral winds set up by the high-latitude convection, we will focus on the polar cap and auroral zone in the rest of the paper. [] Figure illustrates the eastward and southward electric field differences between Runs W and NW, showing the contribution of the neutral winds to the electric field. This contribution will be denoted hereinafter the wind-induced electric field. The dashed lines are for the small source density case of Peymirat et al. [99] (referred to as Run n), and the solid lines are for the present large density case (referred to as Run n). As noted in section, the dynamo effect of the ion-drag-accelerated winds tends to enhance the strength of the electric field and ion convection when the magnetosphere behaves like a current generator, as is approximately the case where Region currents flow. In this picture the zonal winds at auroral latitudes should tend to enhance the meridional electric field, and the meridional winds should tend to enhance the east-west electric field. It turns out that these two tendencies are in partial conflict with each other when account is taken of the constraints that the electric field is curl-free and that the windinduced eastward electric field at the PCB must vanish, since the polar cap potential generator does not allow the wind to modify the potential there. The curl-free condition on the electric field is Rcosl mf cosl m m ¼ Rcosl m : Since cos l m varies relatively little across the auroral zone, in comparison with E ml, () basically states that the southward ðþ ðþ ðþ

4 SMP - PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING Pressure With Winds Pressure Difference: Winds No Winds Max:.9 / Min: MLT Neutral winds at 5 km Max: 7. / Min: 5. MLT Potential With Winds Potential Difference: Winds No Winds Max:.95 / Min:.95 MLT 7m/s MLT Max:.5 / Min:. MLT Figure. SWM voltage generator with a polar cap potential drop of kv and a magnetospheric plasma source density of. cm. The thin solid lines correspond to positive values, the dashed lines correspond to negative values, and the bold line corresponds to zero values. The maximal and minimal values are indicated on the left axis. (top left) Magnetospheric ion pressure (in ppa) with a contour interval of ppa in the equatorial plane of the magnetosphere between the Earth and the dashed circle at. R E, corresponding to the equatorial projection of the polar cap boundary. (top right) Pressure difference induced by the neutral winds (see text) in the equatorial plane of the magnetosphere with a contour interval of 5 ppa. (bottom left) Electric potential (in kv) with a contour interval of kv in the ionosphere between the north pole and invariant latitude, with the magnetic local time indicated on the larger circle. (bottom right) Electric potential induced by the neutral winds (see text) in the ionosphere with a contour interval of.5 kv. (center) Distribution of the neutral winds at an altitude of 5 km. gradient of the eastward electric field equals the eastward gradient of the southward electric field. The eastward and westward winds in the morning and evening tend to generate southward and northward electric fields, respectively. The longitudinal gradient of these meridional electric fields would imply that, at night, the eastward wind-induced electric field would increase as one moves equatorward away from the PCB, where it is zero, and that, during the day, the westward electric field would similarly increase as one moves equatorward. However, an eastward electric field at night and a westward field during the day are in the opposite direction to what the meridional winds tend to generate. Consequently, the resultant wind-induced electric fields cannot simply enhance the convection electric field everywhere in the auroral region. [7] The effects of the constraints on the wind-induced electric field can be seen in Figure. At 9.7 magnetic latitude the eastward wind-induced electric fields for both Runs n and n are small because of proximity to the PCB, where they are constrained to vanish. The small-scale structure is mostly an artifact due to the coarse geographical grid used in the MTIE-GCM to calculate the auroral conductances. At this latitude a slight tendency for an eastward field during the day and a westward field at night is apparent in the direction that the meridional wind should tend to generate. This sense of latitudinal gradient of the zonal windinduced electric field implies that the average eastward gradient of the southward wind-induced field between 9.7 and 7. should be positive during the day and negative at night, as can be detected at 7. latitude in Figure. Note, however, that these longitudinal (or local-time) gradients in the southward electric field are both opposite to those in the total electric field (not shown) and contrary to what one might expect the zonal wind to generate, a consequence of the conflict described above. It is only as one moves further equatorward, beyond 9.7, that the southward windinduced electric field develops a dawn maximum and dusk minimum, consistent with the zonal wind directions and reinforcing the total meridional electric field. This implies, from the curl-free condition, an increase in the eastward wind-induced field at night and in the westward field at day as one moves to lower latitudes, and these increases are indeed seen in Figure. They are contrary to the direction of zonal electric field that the meridional winds tend to generate, again a consequence of the conflict noted above. The wind-induced zonal electric field is also opposite to the total zonal electric field set up by the high-latitude convection (not shown) and tends to reduce the penetration to low and midlatitudes of the zonal electric field around AZEQ. Therefore, with the constraint that the electric field is curl-free, the neutral winds increase the meridional electric field in the auroral zone, decrease the zonal electric field around AZEQ, and enhance the shielding. [] Now let us compare Run n with Run n. Figure illustrates the difference between the neutral winds for Run n and Run n at the same UT, hours after the beginning of the

5 PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING SMP - 5 Eastward Electric Field in mv/m Invariant Latitude Southward Electric Field in mv/m Invariant Latitude Figure. Electric fields as a function of invariant latitude between and the polar cap boundary at 7 for different magnetic local times, as indicated in the lower left of each frame. The potential is imposed in the polar cap with a polar cap potential drop of kv, and the magnetospheric plasma source density is set to. cm. Solid lines are for the run including the ionospheric wind dynamo at all latitudes, dashed lines are for the run with the wind dynamo canceled below the equatorial boundary of the auroral zone at. (indicated by the vertical line), and dotted lines are for the run with no wind dynamo effects. (left) Eastward electric field. (right) Equatorward electric field.

6 SMP - PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING Wind Eastward Electric Field in mv/m Magnetic Local Time Wind Southward Electric Field in mv/m Magnetic Local Time Figure. Electric fields induced by neutral winds (see text) as a function of the magnetic local time for different invariant latitudes associated with the model grid, as indicated in the lower left of each frame. The polar cap boundary is set to 7.. Above this boundary the total potential is set to the Heelis et al. [9] pattern, with a polar cap potential drop of kv. Solid lines are for a magnetospheric plasma source density set to. cm, while dashed lines are for a source density of. cm. (left) Eastward electric field. (right) Equatorward electric field.

7 PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING SMP - 7 m/s MLT Case : kv / Large Density Small Density Figure. Difference between the neutral wind distributions of the large. cm source density run and the small. cm source density run of Peymirat et al. [99]. simulation. The larger magnetospheric plasma density in Run n enhances the shielding associated with the Region field-aligned currents and therefore enhances the amplitudes of the meridional electric field, the zonal ion velocity, and the zonal ion drag on air in the auroral zone. The ion drag is also enhanced by increased electron precipitation into the ionosphere, which increases the electron density. The stronger zonal winds at auroral latitudes increase the wind effect on the shielding, producing a larger reduction of the magnetospheric plasma density close to the Earth. [9] We performed another run with the large. cm source density but a polar cap potential drop of 7 kv. The fractional effect of the neutral wind on the potential and plasma pressure was similar to the Run n simulation described above. [] Let us summarize the results of this section. The influences of the zonal and meridional winds on the auroral-zone electric fields tend partly to oppose each other when the magnetosphere is treated as a potential generator over the polar cap. An increase of the density of the magnetospheric plasma source magnifies the shielding effect induced by the neutral winds. This wind-induced shielding is strong enough to produce a decrease of the plasma pressure by % in the night sector close to the Earth. The relative influence of the neutral winds on shielding does not seem to depend strongly on the magnitude of the polar cap potential drop. The effect of the neutral winds set up by the high-latitude convection is largest in the auroral zone, while the contribution of the low-latitude and midlatitude neutral winds cannot be neglected around the equatorial boundary of the auroral zone.. SWM Hybrid Current/Voltage Generator [] The equation of the electric potential solved by the MTIE-GCM is as given by Richmond T þ m cosl m j " lf þ T m m ¼ mf m m ml cosl mþ R sin Im N cosl m j D N J k N þ sin I m S D S Jk S ; ()

8 SMP - PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING Pressure With Winds Pressure Difference: Winds No Winds Max: 7. / Min: MLT Neutral winds at 5 km Max:.5 / Min: 9.5 MLT Potential With Winds Potential Difference: Winds No Winds Max:. / Min:.5 MLT 77m/s MLT Max:.5 / Min:.757 MLT Figure 5. Same as Figure but for the SWM hybrid current/voltage generator (see text). where T ff, T fl, T lf, and T ll are the ionospheric conductances, K DT mf and K DT ml are the current components driven by the neutral winds, J N and J S are the field-aligned current densities at the top N of the ionosphere in the Northern and Southern Hemispheres, I m and I S m are the northern and southern equivalent dipole inclination angles defined in terms of l m (positive in the north and negative in the south), and D N and D S are scale parameters for the modified apex coordinates, which would be unity in a dipolar magnetic field. The and K D m are magnetic-field-line-integrated quantities, summed over both hemispheres, that also involve coordinatedependent scaling factors. In this equation, is assumed to be symmetric about the magnetic equator, i.e., a function of l m and f m. Any asymmetric component of, as might arise in the polar caps in association with IMF B y effects, for example, must be treated separately; such effects are ignored in the present study. The equatorial boundary condition on requires that the poleward current density summed over the two hemispheres vanishes: T T Rcosl m m j þ KDT ml ¼ : [] If we knew the distribution of magnetospheric field-aligned current into the ionosphere, summed over the two hemispheres, as represented by the second term on the right-hand side of (), we could solve for everywhere over the hemisphere. If, instead, we knew the distribution of electric potential over the high-latitude region down to the AZEQ, below which no magnetospheric current sources exist, then we could solve for at all lower latitudes, using the known distribution at AZEQ as a high-latitude boundary condition. The TIE-GCM uses a hybrid procedure to represent magnetospheric effects: it expresses the field-aligned currents in ð5þ terms of the difference between and a specified reference potential H ( l m, f m )as sin Im N D N J k N þ sin I m S D S Jk S P T ff >< P cos l m ¼ >: l m f m þ 9 T ll cos l m w ½ fh fš>= R ; cos l m >; where P( l m, f m ) is a specified function that controls the relative importance of the ionospheric dynamo and of the reference potential H ; l m and f m are the spatial grid increments in latitude and longitude, respectively; and w is a constant scaling parameter. In the original TIE-GCM, P was set to above the PCB, thereby forcing = H ; it was set to below AZEQ, which lay 5 equatorward of the PCB, and it varied linearly in latitude between these two boundaries. The value of w was set to the latitude difference between the PCB and the AZEQ, expressed in radians. The functional form of the right-hand side of () was chosen for computer-coding convenience, not for any particular geophysical considerations. [] In the MTIE-GCM we use () only above the PCB, which lies at 7.. Below this boundary the magnetospheric source of field-aligned currents is determined by the IMM. For the simulations of the previous section, P = over the polar cap, i.e., is forced to equal H at and above the PCB. In this section we explore the consequences of making P nonzero. When < P <, the magnitude of the field-aligned current is controlled by the difference between the reference potential H and the actual potential. We could eliminate the dependence on, and create a pure current generator, by letting P go toward while simultaneously letting H become ðþ

9 PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING SMP - 9 Wind Eastward Electric Field in mv/m Magnetic Local Time Wind Southward Electric Field in mv/m Magnetic Local Time Figure.. cm. Same as Figure but for the SWM hybrid current/voltage generator (see text). The source density is very large, such that ( P) H remains finite. However, we are interested in exploring a case where the SWM dynamo is intermediate between a pure voltage generator and a pure current generator and where changes in the ionospheric conditions are capable of influencing both the potential and the field-aligned currents over the polar cap, as was found, for example, in MHD simulations by Fedder and Lyons [97]. These authors showed that the SWM dynamo tends to act more like a current generator than like a potential generator, in that changes in the ionospheric conductance produce a larger change in the potential than in the field-aligned current. This change is controlled by the inverse of the effective conductance, which is proportional to the ratio of the magnetic

10 SMP - PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING 5m/s MLT Case :P=.5 Case : P= Figure 7. Difference between the neutral wind distributions for the SWM hybrid current/voltage-generator case and the corresponding SWM voltage generator case (see text). field inside and outside of the magnetopause and to the inverse of the Alfvén speed in the dayside magnetosheath [Hill et al., 97]. We set the parameter P to the constant value.5 inside the polar cap. To have approximately the same potential drop along the PCB as in the simulations of the previous section, we need to double H. The resultant distribution of the potential within the polar cap no longer has the same form as that of H. In particular, the potential extrema now occur within the polar cap instead of at its boundary. (Note that for the potential generator, infinite field-aligned current densities are allowed at the PCB if there is a discontinuity in the meridional electric field there. For the hybrid generator, on the other hand, the Region currents must be spread out in latitude, and discontinuous meridional electric fields are not allowed, meaning that the potential extrema can move away from the PCB.) [] Figure 5 displays in a format similar to Figure the distributions of the pressure and potential for Run W, the distributions of the pressure and potential differences between Runs W and NW, and the distribution of the neutral winds at 5 km. The pressure reaches a value of ppa and the difference pressure reaches a negative value of about 55 ppa on the nightside, corresponding to a relative decrease of 55 ¼ :, very similar to Figure. The only minor difference is the small relative increase of ¼ : on the nightside far away from the Earth. Along the PCB the total potential drop is kv, with a maximum of 7 kv on the dawnside and a minimum of kv on the duskside. The total polar cap potential drop is kv. As we might have anticipated, the winds increase the magnitude of the potential along the PCB, adding. kv on the dawnside and about.5 kv on the duskside, on the order of. the potential drop across the PCB. This implies an increased convection which results in the small positive peak of the difference pressure previously mentioned. [5] Figure shows the wind-induced changes in the eastward and southward electric field at various latitudes. These represent the differences between Runs W and NW. Above 7. these windinduced changes mimic the distributions expected from a twovortex potential pattern with centers located inside the polar cap. Therefore the neutral winds tend to increase the amplitudes of both components of the electric field close to the PCB. Unlike the voltage-generator case, the wind is able to produce zonal electric fields near the PCB that enhance the meridional ionospheric convection. Because these zonal fields decrease rapidly with decreasing latitude away from the PCB, the curl-free constraint

11 PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING SMP - on the total electric field now allows much larger local-time gradients of the southward wind-induced electric field at latitudes poleward of 7, as compared with the voltage-generator case. At latitudes of 7. and below, the direction of the wind-induced zonal electric field reverses, such that it reduces the total zonal electric field. Thus the neutral winds enhance the shielding effect in this case, just as in the cases of the previous section. The alteration of the potential pattern within the polar cap, as compared with the Heelis pattern, allows stronger winds to be produced in the auroral zone. Figure 7 shows the difference between the winds in Figures 5 and. The neutral winds are stronger because the total potential drop inside the polar cap is larger by a factor of.5, and the region where ion-drag forcing is directed in the sense of the auroral winds is considerably broader, extending several degrees of latitude inside the polar cap. This allows stronger east-west winds just outside the polar cap that in turn increase the wind-induced auroral potential drop. Figure 5 allows us to quantify the magnitude of this potential drop. At. magnetic latitude, which we take to represent AZEQ, the potential change due to winds reaches. kv on the duskside and kv on the dawnside. The wind-induced potential drop between the AZEQ and PCB is thus 5.7 kv on the duskside and. kv on the dawnside, and the total normalized wind effect is 5:7þ: ¼ : [] In order to compare directly the SWM hybrid current/ voltage generator with a SWM voltage generator that has a similar potential distribution over the polar cap, we perform a further simulation. We take the polar cap potential pattern computed after hours in the previous simulation with P =.5, neglecting the wind dynamo, and keep it fixed inside the polar cap. Although the potential at the PCB is not allowed to change in this simulation, unlike the hybrid current/voltage generator where it does change, we find that the wind-induced potentials for these two simulations are nearly the same at the AZEQ. The wind-induced potential drops in the dawn and dusk sectors across the auroral zone are smaller for this voltage-generator case than for the hybrid current/ voltage-generator case, even though the wind velocity is only slightly reduced in this case. We are encountering the same effect described in the previous section: when the wind cannot change the potential at the PCB, there is a tendency for the dynamo effects of the meridional winds partially to offset those of the zonal winds. [7] In summary, our hybrid model allows winds to increase the potential in the polar cap and around the PCB. However, the increased wind-induced potential drop at the PCB in the hybrid model does not show up as an increased zonal electric field at the equatorward edge of the auroral zone, as hypothesized by Peymirat et al. [99], because its influence is offset by the increased windinduced latitudinal potential drops across the auroral zone. 5. Conclusions [] The MTIE-GCM of Peymirat et al. [99] is used to study the steady state electrodynamical coupling of the thermosphere, the ionosphere, and the magnetosphere for cases where the SWM dynamo acts either as a voltage generator or as a hybrid current/ voltage generator over the polar cap. [9] A detailed examination of the electric field distribution illustrates the way the neutral winds accelerated by ion drag act on the electric field. To the extent the magnetosphere acts as a current generator, the neutral winds tend to generate electric fields that produce ion convection entrained with the wind velocity, subject to the constraints that the electric field be constant with height and curlfree. Boundary conditions can impose additional constraints. For the SWM voltage generator the potential at the PCB is fixed, which is found to restrict the ability of ion-drag-driven winds to modulate the potential within the auroral zone and the AZEQ. The high-latitude neutral winds tend to increase the meridional electric field at highlatitudes but to decrease the zonal electric field at AZEQ, which corresponds to enhanced shielding. The respective tendencies of the meridional and zonal winds to modulate the zonal electric field around AZEQ are in partial conflict with the meridional wind tending to increase the total zonal electric field, and thus to reduce shielding, and the zonal wind tending to decrease the zonal field, and thus to enhance shielding. Nonetheless, the net influence of the winds is to enhance the shielding. For a source density of. cm and a polar cap potential drop of kv, the ionospheric dynamo induced by the neutral winds reduces the potential variation around the AZEQ by % and depletes the nightside magnetosphere by %. The effect of the neutral winds on the electric field is due to neutral winds of all latitudes. The high-latitude neutral winds act mainly on the auroral electric field and have minor influence equatorward of the auroral zone where the low and midlatitude neutral winds start to act. A doubling of the magnetospheric plasma source density amplifies the Region field-aligned currents due to the MT dynamo and also amplifies particle precipitation, leading to greater auroral-zone electron densities and currents. The stronger ion drag induces larger zonal winds that reinforce the shielding. The wind-induced reduction in the nightside magnetospheric plasma pressure is proportionately two times greater than the increase in the plasma source density, while the wind-induced change in potential around the AZEQ is changed only little, indicating significant nonlinear behavior. [] An increase of the polar cap potential drop up to 7 kv does not change the relative importance of the neutral wind effect on shielding. [] When the SWM dynamo is treated as a hybrid current/ voltage generator over the polar cap, the resulting electric potential and magnetospheric plasma convection are influenced differently by winds than for the case of a pure SWM voltage generator. For the hybrid generator the neutral winds are allowed to amplify the potential variation around the PCB, by %, and the potential drops across the morning and evening auroral zones, by about %. The amplified westward electric field along the PCB on the nightside results in a small wind-induced increase in the earthward motion of the plasma and a small increase of the magnetospheric pressure by % far away from the Earth. At night the larger eastward gradient of the equatorward wind-induced electric field for the hybridgenerator case is accompanied by a larger equatorward gradient of the eastward wind-induced field owing to the curl-free constraint, such that there is practically no change in the wind-induced potential around the AZEQ. Thus the hypothesis suggested by Peymirat et al. [99], that allowing the winds to vary the potential along the PCB could decrease the contribution of ion-drag-driven winds to the shielding, is not supported by these simulations. [] Acknowledgments. This study was supported by the NASA Sun-Earth Connection Theory Program. [] Michel Blanc thanks Stanislav Sazykin and another referee for their assistance in evaluating this paper. References Axford, W. I., and C. O. Hines, A unifying theory of high-latitude geophysical phenomena and geomagnetic storms, Can. J. Phys., 9,, 9. Banks, P. M., Magnetosphere processes and the behavior of the neutral atmosphere, Space Res.,, 5 7, 97. Baumjohann, W., and G. Paschmann, Determination of the polytropic index in the plasma sheet, Geophys. Res. Lett.,, 95 9, 99. Baumjohann, W., G. Paschmann, and C. A. Cattell, Average plasma properties in the central plasma sheet, J. Geophys. Res., 9, 597, 99. Block, L. P., On the distribution of electric fields in the magnetosphere, J. Geophys. Res, 7,, 9. Boyle, C. B., P. H. Reiff, and M. R. Hairston, Empirical polar cap potentials, J. Geophys. Res.,, 5, 997. Deng, W., T. L. Killeen, and A. G. Burns, The flywheel effect: Ionospheric currents after a geomagnetic storm, Geophys. Res. Lett.,, 5, 99. Deng, W., T. L. Killeen, G. Burns, and R. G. Roble, The effects of neutral inertia on ionospheric currents in the high-latitude thermosphere following a geomagnetic storm, J. Geophys. Res., 9, , 99.

12 SMP - PEYMIRAT ET AL.: WIND-MAGNETOSPHERE IONOSPHERE COUPLING Escoubet, C. P., A. Pedersen, R. Schmidt, and P. A. Lindqvits, Density in the magnetosphere inferred from ISEE spacecraft potential, J. Geophys. Res.,, 7,595 7,9, 997. Fedder, J. A., and J. G. Lyons, The solar wind-magnetosphere-ionosphere current-voltage relationship, Geophys. Res. Lett.,,, 97. Fejer, J., Theory of the geomagnetic daily disturbances variations, J. Geophys. Res, 9, 7, 9. Fejer, B. G., R. W. Spiro, R. A. Wolf, and J. C. Foster, Latitudinal variation of perturbation electric fields during magnetically disturbed periods: 9 SUNDIAL observations and model results, Ann. Geophys.,, 5, 99. Forbes, J. M., and M. Harel, Magnetosphere-Thermosphere coupling: An experiment in interactive modeling, J. Geophys. Res., 9,, 99. Goertz, C. K., and W. Baumjohann, On the thermodynamics of the plasma sheet, J. Geophys. Res., 9,,99,99, 99. Hardy, D. A., M. S. Gussenhoven, and E. Holeman, A statistical model of auroral electron precipitation, J. Geophys. Res., 9, 9, 95. Hardy, D. A., M. S. Gussenhoven, and D. Brautigam, A statistical model of auroral ion precipitation, J. Geophys. Res., 9, 7 9, 99. Harel, M., R. A. Wolf, P. H. Reiff, R. W. Spiro, W. J. Burke, F. J. Rich, and M. Smiddy, Quantitative simulation of a magnetospheric substorm,, Model logic and overview, J. Geophys. Res.,, 7, 9. Heelis, R. A., J. K. Lowell, and R. W. Spiro, A model of the high-latitude ionospheric convection pattern, J. Geophys. Res., 7, 9 5, 9. Heinemann, M., Role of collisionless heat flux in magnetospheric convection, J. Geophys. Res.,,,,97, 999. Hill, T. W., A. J. Dessler, and R. A. Wolf, Mercury and Mars: The role of ionospheric conductivity in the acceleration of magnetospheric particles, Geophys. Res. Lett.,, 9, 97. Huang, C. Y., and L. A. Frank, A statistical study of the central plasma sheet: Implications for substorm models, Geophys. Res. Lett.,, 5 55, 9. Huang, C. Y., and L. A. Frank, A statistical survey of the central plasma sheet, J. Geophys. Res., 99, 95, 99. Huang, C. Y., C. K. Goertz, L. A. Frank, and G. Rostoker, Observational determination of the adiabatic index in the quiet time plasma sheet, Geophys. Res. Lett.,, 5 5, 99. Lu, G., A. D. Richmond, B. A. Emery, and R. G. Roble, Magnetosphereionosphere-thermosphere coupling: Effect of neutral winds on energy transfer and field-aligned current, J. Geophys. Res.,, 9, 9,59, 995. Lyons, L. R., T. L. Killeen, and R. L. Walterscheid, The neutral wind flywheel as a source of quiet-time, polar cap currents, Geophys. Res. Lett.,,, 95. Peymirat, C., and D. Fontaine, Numerical simulation of magnetospheric convection including the effect of field-aligned currents and electron precipitation, J. Geophys. Res., 99,,55,7, 99. Peymirat, C., and D. Fontaine, Polar cap convection patterns inferred from EISCAT observations, Ann. Geophys., 5,, 997. Peymirat, C., A. D. Richmond, B. A. Emery, and R. G. Roble, A magnetosphere-thermosphere-ionosphere electrodynamics general circulation model, J. Geophys. Res.,, 7,7 7,77, 99. Richmond, A. D., Ionospheric electrodynamics using magnetic apex coordinates, J. Geomagn. Geoelectr., 7, 9, 995a. Richmond, A. D., The ionospheric wind dynamo: Effects of its coupling with different atmospheric regions, in The Upper Mesosphere and Lower Thermosphere: A Review of Experiment and Theory, Geophys. Monogr. Ser., vol. 7, edited by R. M. Johnson, and T. L. Killeen, pp. 9, AGU, Washington, D. C., 995b. Richmond, A. D., and S. Matsushita, Thermospheric response to a magnetic substorm, J. Geophys. Res.,, 9 5, 975. Richmond, A. D., E. C. Ridley, and R. G. Roble, A thermosphere/ionosphere general circulation model with coupled electrodynamics, Geophys. Res. Lett., 9,, 99. Schumaker, T. L., M. S. Gussenhoven, D. A. Hardy, and R. L. Carovillano, The relationship between diffuse auroral and plasma sheet electron distributions near local midnight, J. Geophys. Res, 9,,,7, 99. Spiro, R. W., R. A. Wolf, and B. G. Fejer, Penetration of high-latitudeelectric-field effects to low latitudes during SUNDIAL 9, Ann. Geophys.,, 9, 9. Vasyliunas, V. M., Mathematical models of magnetospheric convection and its coupling to the ionosphere, in Particles and Fields in the Magnetosphere, edited by M. McCormac, pp. 7, D. Reidel, Norwell, Mass., 97. Vasyliunas, V. M., The interrelationship of magnetospheric processes, in Earth s Magnetosphere Processes, edited by M. McCormac, pp. 9, D. Reidel, Norwell, Mass., 97. Wolf, R. A., G. A. Mantjoukis, and R. W. Spiro, Theoretical comments on the nature of the plasmapause, Adv. Space Res., (), 77, 9. C. Peymirat, Centre d Etude Spatiale des Rayonnements, 9 Avenue du Colonel Roche, Toulouse Cedex, France (peymirat@cesr.fr) A. D. Richmond and R. G. Roble, High Altitude Observatory, National Center for Atmospheric Research, Boulder, CO 7-, USA. (richmond@ucar.edu; roble@ucar.edu)